Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    INDEX
    Explanations

    textual patterns that are likely noise or errors, as they consist of random or nonsensical characters

    oai_token-act-pair · gpt-3.5-turbo

    special characters and variations of "DI" potentially related to identification or classification

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Negative Logits
    iage
    -0.88
     Leopard
    -0.73
    engers
    -0.73
     Lans
    -0.72
     Seah
    -0.72
    mares
    -0.72
     Hole
    -0.69
    itudinal
    -0.67
     Bengal
    -0.66
    oir
    -0.66
    POSITIVE LOGITS
    âĸijâĸij
    1.20
    女
    1.19
    ption
    1.04
    actus
    1.03
    entric
    1.00
    andom
    0.94
    LECT
    0.93
    çĶŁ
    0.92
    éĹ
    0.90
    âĸij
    0.88
    Act Density 0.025%

    No Known Activations